HippoBaro commented on code in PR #9795:
URL: https://github.com/apache/arrow-rs/pull/9795#discussion_r3134772233


##########
parquet/src/column/writer/mod.rs:
##########
@@ -708,15 +694,15 @@ impl<'a, E: ColumnValueEncoder> GenericColumnWriter<'a, 
E> {
                 ));
             }
 
-            // Count the occasions where we start a new row
-            for &level in levels {
-                self.page_metrics.num_buffered_rows += (level == 0) as u32
-            }
-
-            // Update histogram
-            self.page_metrics.update_repetition_level_histogram(levels);
-
-            self.rep_levels_encoder.put(levels);
+            let mut new_rows = 0u32;
+            self.rep_levels_encoder
+                .put_with_observer(levels, |level, count| {
+                    new_rows += (count as u32) * (level == 0) as u32;
+                    if let Some(ref mut h) = 
self.page_metrics.repetition_level_histogram {

Review Comment:
   Just did the experiment locally with this:
   
   ```rust
   match self.page_metrics.definition_level_histogram.as_mut() {
       Some(histogram) => encoder.put_with_observer(levels, |level, count| {
           values_to_write += count * (level == max_def) as usize;
           histogram.increment_by(level, count as i64);
       }),
       None => encoder.put_with_observer(levels, |level, count| {
           values_to_write += count * (level == max_def) as usize;
       }),
   };
   ```
   
   Benchmarks show a 2-3% improvements on average. 



##########
parquet/src/column/writer/mod.rs:
##########
@@ -708,15 +694,15 @@ impl<'a, E: ColumnValueEncoder> GenericColumnWriter<'a, 
E> {
                 ));
             }
 
-            // Count the occasions where we start a new row
-            for &level in levels {
-                self.page_metrics.num_buffered_rows += (level == 0) as u32
-            }
-
-            // Update histogram
-            self.page_metrics.update_repetition_level_histogram(levels);
-
-            self.rep_levels_encoder.put(levels);
+            let mut new_rows = 0u32;
+            self.rep_levels_encoder
+                .put_with_observer(levels, |level, count| {
+                    new_rows += (count as u32) * (level == 0) as u32;
+                    if let Some(ref mut h) = 
self.page_metrics.repetition_level_histogram {

Review Comment:
   Just did the experiment locally with this:
   ```rust
   match self.page_metrics.definition_level_histogram.as_mut() {
       Some(histogram) => encoder.put_with_observer(levels, |level, count| {
           values_to_write += count * (level == max_def) as usize;
           histogram.increment_by(level, count as i64);
       }),
       None => encoder.put_with_observer(levels, |level, count| {
           values_to_write += count * (level == max_def) as usize;
       }),
   };
   ```
   
   Benchmarks show a 2-3% improvements on average. 



-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to